CN108052952A - A kind of the clothes similarity determination method and its system of feature based extraction - Google Patents
A kind of the clothes similarity determination method and its system of feature based extraction Download PDFInfo
- Publication number
- CN108052952A CN108052952A CN201711377528.0A CN201711377528A CN108052952A CN 108052952 A CN108052952 A CN 108052952A CN 201711377528 A CN201711377528 A CN 201711377528A CN 108052952 A CN108052952 A CN 108052952A
- Authority
- CN
- China
- Prior art keywords
- similarity
- clothes
- image
- feature
- clothing
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
- G06V10/751—Comparing pixel values or logical combinations thereof, or feature values having positional relevance, e.g. template matching
Abstract
The embodiment of the invention discloses the clothes similarity determination method and its system of a kind of extraction of feature based, wherein, this method includes：Obtain the image data for carrying out clothes similarity judgement；The upload and download feature of clothes is parsed, carries out image of clothing feature extraction；The judgement of clothes similarity is carried out according to the image of input, provides level of similarity.In embodiments of the present invention, establish the analytic method of the upload and download of clothes, the key feature for influencing clothes similarity is demarcated, and establish the deep learning network under the more attributes of clothes, and then the similarity carried out to clothes under multiple features judges, by the degree of similarity of measuring similarity data characterization clothes, personalized clothes selection and recommendation function are more in line with.
Description
Technical field
The present invention relates to the clothes similarities that computer image processing technology field more particularly to a kind of feature based are extracted
Determination method and its system.
Background technology
With the promotion of people's quality of the life, the selection of clothes is pursued there has also been more.The proper clothes of fashion can allow
User is walked in the forward position of fashion, shows personal taste, user is helped to establish confidence, while clothes can also reflect the personality of itself
Feature and hidden feeling etc..How satisfied clothes are found in substantial amounts of toggery, be computer image processing technology
Popular problem, has a wide range of applications meaning.The present invention is just intended to, by carrying out character representation to image of clothing, establish clothes
The quantitative approach of similarity, the application demand searched so as to fulfill clothes similarity.
Clothes have complicated structure, and abundant minutia causes clothes to have complicated and abundant style variation.Meter
The technology of calculation machine image procossing has in terms of image identification, image retrieval to be widely applied very much.For the processing of image of clothing,
Good effect can be obtained in terms of image segmentation, image denoising and image enhancement, it is grand in terms of the characteristic processing of image
Basic description can also be obtained in sight.But in the processing of clothes minutia, there are no the requirements for reaching application, this is directly
The lookup of clothes is affected, that is, in terms of clothes similarity processing, the requirement of application can't be met well.
In order to carry out clothes similarity quantitative judgement, it is necessary to image of clothing carry out deeper into processing.Existing skill
Art has relatively good for the rough sort of image of clothing as a result, still minutia also directly influences the accuracy of lookup.Cause
This is analyzed and is handled to the minutia of clothes, is established feature tag for image of clothing, and is established and taken according to feature tag
Searching database is filled, the similar grade for clothes minutia can be not only obtained, image of clothing can also be provided to the user
Fine lookup, meet popular, fashion and the application demand of personalization.
The content of the invention
It is an object of the invention to overcome the deficiencies in the prior art, and the present invention provides a kind of clothes of feature based extraction
Similarity determination method and its system establish the deep learning network under the more attributes of clothes, and then carry out multiple features to clothes
Under similarity judge, by the degree of similarity of measuring similarity data characterization clothes, be more in line with personalized clothes choosing
It selects and recommendation function.
To solve the above-mentioned problems, the present invention proposes a kind of clothes similarity determination method of feature based extraction, institute
The method of stating includes：
Obtain the image data for carrying out clothes similarity judgement；
The upload and download feature of clothes is parsed, carries out image of clothing feature extraction；
The judgement of clothes similarity is carried out according to the image of input, provides level of similarity.
Preferably, described the step of obtaining the image data for carrying out clothes similarity judgement, including：
Obtain the image of clothing for carrying out similarity judgement；
Location determination is carried out for image of clothing；
Establish the location parameter of image of clothing..
Preferably, the step of upload and download feature of the parsing clothes, progress image of clothing feature extraction, including：
According to the location parameter of image of clothing, the upper dress part of image and lower dress part are determined；
According to the characteristics of toggery, calibration carries out the characteristic parameter of similarity judgement；
According to the garment feature parameter of calibration, multi-tag model is established；
It establishes garment feature and represents database；
Training multi-tag model, establishes similarity decision model.
Preferably, the garment feature parameter according to calibration, the step of establishing multi-tag model, including：
Establish study and training that deep learning training pattern carries out garment feature parameter；
Garment feature extraction process is carried out according to deep learning training pattern；
Establish a kind of 50 minutia labels of clothes attribute.
Preferably, the judgement that clothes similarity is carried out according to the image of input, the step of providing level of similarity, bag
It includes：
Feature extraction is carried out to the image of clothing of acquisition；
Similarity judgement is carried out by image in multi-tag model and database；
Comprehensive analysis judges as a result, the order successively decreased by similarity provides the high clothes series of similarity.
Correspondingly, the present invention also provides a kind of feature based extraction clothes similarity decision-making system, the system comprises：
Image collection module, for obtaining the image of clothing for carrying out similarity judgement；
Characteristic extracting module determines the upper dress part of image and lower dress part；According to the characteristics of toggery, demarcate into
The characteristic parameter that row similarity judges；According to the garment feature parameter of calibration, multi-tag model is established；Establish garment feature expression
Database；
Similarity determination module judges for carrying out the similarity of image of clothing and database data, and provides similarity
Rank.
Preferably, described image acquisition module includes：
Acquiring unit, for obtaining the clothes parameter for carrying out similarity judgement；
First computing unit carries out location determination processing for image of clothing；
Second computing unit establishes the location parameter of image of clothing.
Preferably, the characteristic extracting module includes：
3rd computing unit for given image, calculates the upper dress part of graphical representation and lower dress part；
4th computing unit, according to the characteristics of toggery, calibration carries out the characteristic parameter of similarity judgement；According to depth
It spends learning training model and carries out garment feature extraction process, establish multi-tag model；
Unit is established, the garment feature parameter obtained according to deep learning training pattern establishes garment feature and represents data
Storehouse.
Preferably, the 4th computing unit is additionally operable to according to the multi-tag model obtained, to given image of clothing
It is handled, while garment feature parameter is obtained according to deep learning training pattern, as the judgement for carrying out clothes similarity
Data.
Preferably, the similarity determination module includes：
Analytic unit carries out feature extraction for the image of clothing for carrying out similarity determination processing, establishes garment feature ginseng
Number；
Identifying unit carries out the calibration result of the characteristic of clothes for comprehensive analysis, provides clothes level of similarity.
Implement the embodiment of the present invention, establish the deep learning network under the more attributes of clothes, and then clothes are carried out mostly special
Similarity under sign judges, by the degree of similarity of measuring similarity data characterization clothes, is more in line with personalized clothes
Selection and recommendation function.
Description of the drawings
It in order to illustrate more clearly about the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existing
There is attached drawing needed in technology description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention, for those of ordinary skill in the art, without creative efforts, can be with
Other attached drawings are obtained according to these attached drawings.
Fig. 1 is the flow diagram of the clothes similarity determination method of the feature based extraction of the embodiment of the present invention；
Fig. 2 is the flow diagram that the image data for carrying out clothes similarity judgement is obtained in the embodiment of the present invention；
Fig. 3 is the upload and download feature that clothes are parsed in the embodiment of the present invention, carries out the stream of image of clothing feature extraction
Journey schematic diagram；
Fig. 4 is the flow diagram for establishing multi-tag model in the embodiment of the present invention according to the garment feature parameter of calibration；
Fig. 5 is the judgement for carrying out clothes similarity in the embodiment of the present invention according to the image of input, provides level of similarity
Flow diagram；
Fig. 6 is the structure composition schematic diagram of the clothes similarity decision-making system of the feature based extraction of the embodiment of the present invention.
Specific embodiment
Below in conjunction with the attached drawing in the embodiment of the present invention, the technical solution in the embodiment of the present invention is carried out clear, complete
Site preparation describes, it is clear that described embodiment is only part of the embodiment of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, those of ordinary skill in the art are obtained every other without creative efforts
Embodiment belongs to the scope of protection of the invention.
Fig. 1 be the embodiment of the present invention feature based extraction clothes similarity determination method flow diagram, such as Fig. 1
Shown, this method includes：
S1 obtains the image data for carrying out clothes similarity judgement；
S2 parses the upload and download feature of clothes, carries out image of clothing feature extraction；
S3 carries out the judgement of clothes similarity according to the image of input, provides level of similarity.
Wherein, as shown in Fig. 2, S1 further comprises：
S11 obtains the image of clothing for carrying out similarity judgement；Image of clothing is represented in a manner of bitmap herein；Image
Data can represent the state of human dressing；
S12 carries out location determination for image of clothing, the image data of acquisition is pre-processed, is employed herein
Faster-RCNN networks can quickly choose the personage in image of clothing with bounding box, the model in center
BBox by be area maximum region, become region to be selected；
In S12, the image of input is put into convolutional neural networks handles first by Faster-RCNN；Hereafter use
Window is suggested in RPN (Region Proposal Network) generations, and 300 suggestion windows are generated per pictures；By building for generation
View window is re-mapped on last layer of characteristic pattern of convolutional neural networks；It is raw by pooling layers of each Rol of Rol
Into the characteristic pattern of fixed size；Finally class probability and frame are returned using Softmax Loss and Smooth L1Loss
Return carry out joint training；
S13 establishes the location parameter of image of clothing, is analyzed according to the shooting angle of image, the model's in center
BBox by be area maximum region, all BBox can be then ranked up, retain maximum one as the figure handled
As data, other background parts will be used as noise remove.
Further, as shown in figure 3, S2 includes：
S21 according to the location parameter of image of clothing, determines the upper dress part of image and lower dress part；When one group of input
Image is when being shone comprising the whole body including personage, also uncorrelated since whole body is according to containing above the waist and the clothes of the lower part of the body
Background and header information, the then ratio that refer to human body carry out rough division, obtain individual upper body clothes and be taken with the lower part of the body
Dress；
The ratio with reference to human body carries out what rough division was performed such：
Division obtains the scope of upper body clothes, and vectorial the first two value represents the transverse and longitudinal coordinate in the scope upper left corner, latter two
Value represents the length and height of scope respectively：
[x11+(x21-x11)/6,y11+(y21-y11)/6,(x21-x11)/3*2,(y21-y11)/3]
Similarly, division obtains the scope of lower body garment：
[x11+(x21-x11)/6,(y21+y11)/2,(x21-x11)/3*2,(y21-y11)/2]
The purpose so set is to reduce the scope for including clothes as much as possible, removes incoherent background most possibly
Or the information such as human skin, improve the accuracy of method.
S22, according to the characteristics of toggery, calibration carries out the characteristic parameter of similarity judgement；Wherein style and style
Image feature extraction is mainly carried out by deep learning mode, and color characteristic is then described by color histogram, and most
It is concluded a kind of to certain in 6 kinds of primary colors at last, facilitates the colouring information of description clothes；
S23 according to the garment feature parameter of calibration, establishes multi-tag model；Deep learning and training are carried out, establishes and uses
In the feature tag that similarity judges；Before model training is carried out, it is necessary first to initial data is pre-processed, remove with
The incoherent impurity image of task.Corresponding weak label then is stamped to data set, weak label carries out not in pixel scale
The calibration of label, but only show that corresponding feature whether there is on the image.Mark Pixel-level labeling requirement can so be saved
Plenty of time.Meanwhile when adding new image data, also can soon it be labeled；
S24 establishes garment feature and represents database；In order to carry out similarity judgement, it is necessary first to which establishing has clothes category
The database of property has recorded the relevant information including image of clothing and garment feature in database, and the foundation of this database is
It is completed according to the mark of attribute, the data of database include 32133 image datas；Database is using MySQL.
S25, training multi-tag model establish similarity decision model, and the network model of training employs VGG16 nets
Network, carries out image of clothing up to ten thousand multiple two-value property classification predictions, and trained iterations is arranged to 450,000 times.
In S23, further, as shown in figure 4, S23 includes：
S231 establishes study and training that deep learning training pattern carries out garment feature parameter；In order to realize to clothes
The prediction of image attributes, it is necessary to be repaiied to network structure before when application VGG16 networks carry out deep learning training pattern
The problem of changing, just can be suitably used for the extraction of current clothes attributive character, compared with original network, amended Web vector graphic two
It is data layers a, correspond respectively to label and image data.Network profile needs slice layers of addition to separate label；
S232 carries out garment feature extraction process according to deep learning training pattern；Depth is carried out using VGG16 networks
Practise training pattern, while the step-length that convolution kernel is moved suitably reduces, with adapt to it is some on the image and the spy that less attracts attention
Sign, such as neckline shape.In the last of network structure, for each clothes attribute set up respectively one innerProduct layers, one
It is Accuracy layers a, one SoftmaxWithLoss layers, so train come model can predict the clothes attribute be 0
Or it is 1, that is, judges attribute presence or absence.The model that final training obtains can carry out feature to newly-increased image data and carry
It takes.
S233 establishes a kind of 50 minutia labels of clothes attribute；
In specific implementation, S233 implementation processes are as follows：
The attribute extracted is needed to include upper body clothes length (long, medium and short), coat-sleeve length (sleeveless, short, long), neckline shape
Shape (stand-up collar, V necks, wide shallow neckline, crew neck, lapel, polo-neck, company's cap shirt), the loose degree of clothes (tight, medium, loose), stamp
(pure color, grid, dot, decorative pattern, horizontal stripe, nicking, number or letter, repeatedly pattern), trousers length (long, medium and short), trousers
Sub- stamp (identical with upper body, totally 8 kinds), the loose degree (identical with upper body, totally 3 kinds) of trousers, skirt length (it is long, in,
It is short), the pattern (A types skirt, bag stern skirt) of skirt, the stamp (identical with upper body, totally 8 kinds) of skirt.Attributive character in 51 in total.
Further, as shown in figure 5, S3 includes：
S31 carries out feature extraction to the image of clothing of acquisition；The clothes figure that the model that mainly application training obtains obtains
As data progress feature extraction, image data and label data are converted to LMDB forms respectively, establish minutia；
S32 carries out similarity judgement by image in multi-tag model and database；The clothes of feature tag will be established
Image carries out similarity judgement with the garment data in database；
It in S32, is directly calculated for color attribute using Euclidean distance, after normalized, per two pieces clothes
Color histogram similarity calculate distribution of results and 0~1 between, be more close to 0 it is more similar on color attribute, it is on the contrary then
Otherness is bigger.
It is divided into five kinds of features for style characteristics, by 51 kinds of clothes attributes with clothes length, sleeve length, collar, version type, stamp
Vector calculates similarity respectively.Wherein, length, sleeve length, collar, the similarity calculation of version type may be employed Euclidean distance, because
With uniqueness, final result is only present with 1 or 0 situation.And a clothes can have a variety of stamps simultaneously, so
It needs first to be normalized, then using Euclidean distance；Or directly calculate the cosine similarity of two stamp vectors, final knot
Fruit is between zero and one.
S33, comprehensive analysis judges as a result, the order successively decreased by similarity provides the high clothes series of similarity.Most last phase
Like degree calculation basis：
Similarity value=0.4* color attributes+0.6* (0.2* clothes length+0.2* sleeve lengths+0.1* editions types of+0.1* collars
+ 0.4* stamps)
When similarity value closer to 0 when, clothes similarity degree is higher, and closer to 1, then similarity degree is lower.
Correspondingly, the embodiment of the present invention also provides a kind of clothes similarity decision-making system of feature based extraction, such as Fig. 6 institutes
Show, which includes：
Image collection module 1, for obtaining the image of clothing for carrying out similarity judgement；
Characteristic extracting module 2 determines the upper dress part of image and lower dress part；According to the characteristics of toggery, calibration
Carry out the characteristic parameter of similarity judgement；According to the garment feature parameter of calibration, multi-tag model is established；Establish garment feature table
Show database；
Similarity determination module 3 judges for carrying out the similarity of image of clothing and database data, and provides similarity
Rank.
Image collection module 1 includes：
Acquiring unit, for obtaining the clothes parameter for carrying out similarity judgement；
First computing unit carries out location determination processing for image of clothing；
Second computing unit establishes the location parameter of image of clothing.
Characteristic extracting module 2 includes：
3rd computing unit for given image, calculates the upper dress part of graphical representation and lower dress part；
4th computing unit, according to the characteristics of toggery, calibration carries out the characteristic parameter of similarity judgement；According to depth
It spends learning training model and carries out garment feature extraction process, establish multi-tag model；
Unit is established, the garment feature parameter obtained according to deep learning training pattern establishes garment feature and represents data
Storehouse.
4th computing unit is additionally operable to according to the multi-tag model obtained, at given image of clothing
Reason, while garment feature parameter is obtained according to deep learning training pattern, the data as the judgement for carrying out clothes similarity.
Similarity determination module 3 includes：
Analytic unit carries out feature extraction for the image of clothing for carrying out similarity determination processing, establishes garment feature ginseng
Number；
Identifying unit carries out the calibration result of the characteristic of clothes for comprehensive analysis, provides clothes level of similarity.
The function of each function module can be found at the flow in the method for the present invention embodiment in the system embodiment of the present invention
Reason, which is not described herein again.
Implement the embodiment of the present invention, establish the deep learning network under the more attributes of clothes, and then clothes are carried out mostly special
Similarity under sign judges, by the degree of similarity of measuring similarity data characterization clothes, is more in line with personalized clothes
Selection and recommendation function.
One of ordinary skill in the art will appreciate that all or part of step in the various methods of above-described embodiment is can
Relevant hardware to be instructed to complete by program, which can be stored in a computer readable storage medium, storage
Medium can include：Read-only memory (ROM, Read Only Memory), random access memory (RAM, Random
Access Memory), disk or CD etc..
In addition, the clothes similarity determination method that the feature based provided above the embodiment of the present invention extracts carries out
It is discussed in detail, specific case used herein is set forth the principle of the present invention and embodiment, above example
Illustrate to be only intended to help the method and its core concept for understanding the present invention；Meanwhile for those of ordinary skill in the art, according to
According to the thought of the present invention, there will be changes in specific embodiments and applications, in conclusion this specification content
It should not be construed as limiting the invention.
Claims (10)
1. a kind of clothes similarity determination method of feature based extraction, which is characterized in that the described method includes：
Obtain the image data for carrying out clothes similarity judgement；
The upload and download feature of clothes is parsed, carries out image of clothing feature extraction；
The judgement of clothes similarity is carried out according to the image of input, provides level of similarity.
2. the clothes similarity determination method of feature based as described in claim 1 extraction, which is characterized in that it is described obtain into
The step of image data that row clothes similarity judges, including：
Obtain the image of clothing for carrying out similarity judgement；
Location determination is carried out for image of clothing；
Establish the location parameter of image of clothing.
3. the clothes similarity determination method of feature based extraction as described in claim 1, which is characterized in that the parsing clothes
The step of upload and download feature of dress, progress image of clothing feature extraction, including：
According to the location parameter of image of clothing, the upper dress part of image and lower dress part are determined；
According to the characteristics of toggery, calibration carries out the characteristic parameter of similarity judgement；
According to the garment feature parameter of calibration, multi-tag model is established；
It establishes garment feature and represents database；
Training multi-tag model, establishes similarity decision model.
4. the clothes similarity determination method of feature based extraction as claimed in claim 3, which is characterized in that described according to mark
Fixed garment feature parameter, the step of establishing multi-tag model, including：
Establish study and training that deep learning training pattern carries out garment feature parameter；
Garment feature extraction process is carried out according to deep learning training pattern；
Establish a kind of 50 minutia labels of clothes attribute.
5. the clothes similarity determination method of the feature based extraction as described in claim 1 or 3 or 4, which is characterized in that described
The step of judgement of clothes similarity is carried out according to the image of input, provides level of similarity, including：
Feature extraction is carried out to the image of clothing of acquisition；
Similarity judgement is carried out by image in multi-tag model and database；
Comprehensive analysis judges as a result, the order successively decreased by similarity provides the high clothes series of similarity.
6. a kind of clothes similarity decision-making system of feature based extraction, which is characterized in that the system comprises：
Image collection module, for obtaining the image of clothing for carrying out similarity judgement；
Characteristic extracting module determines the upper dress part of image and lower dress part；According to the characteristics of toggery, calibration carries out phase
The characteristic parameter judged like degree；According to the garment feature parameter of calibration, multi-tag model is established；It establishes garment feature and represents data
Storehouse；
Similarity determination module judges for carrying out the similarity of image of clothing and database data, and provides level of similarity.
7. the clothes similarity decision-making system of feature based extraction as claimed in claim 6, which is characterized in that described image obtains
Modulus block includes：
Acquiring unit, for obtaining the clothes parameter for carrying out similarity judgement；
First computing unit carries out location determination processing for image of clothing；
Second computing unit establishes the location parameter of image of clothing.
8. the clothes similarity decision-making system of feature based extraction as claimed in claim 6, which is characterized in that the feature carries
Modulus block includes：
3rd computing unit for given image, calculates the upper dress part of graphical representation and lower dress part；
4th computing unit, according to the characteristics of toggery, calibration carries out the characteristic parameter of similarity judgement；According to depth
It practises training pattern and carries out garment feature extraction process, establish multi-tag model；
Unit is established, the garment feature parameter obtained according to deep learning training pattern establishes garment feature and represents database.
9. the clothes similarity decision-making system of feature based as claimed in claim 7 or 8 extraction, which is characterized in that described the
Four computing units are additionally operable to according to the multi-tag model obtained, and given image of clothing is handled, while according to depth
Learning training model obtains garment feature parameter, the data as the judgement for carrying out clothes similarity.
10. the clothes similarity decision-making system of feature based extraction as claimed in claim 6, which is characterized in that described similar
Degree determination module includes：
Analytic unit carries out feature extraction for the image of clothing for carrying out similarity determination processing, establishes garment feature parameter；
Identifying unit carries out the calibration result of the characteristic of clothes for comprehensive analysis, provides clothes level of similarity.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711377528.0A CN108052952A (en) | 2017-12-19 | 2017-12-19 | A kind of the clothes similarity determination method and its system of feature based extraction |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711377528.0A CN108052952A (en) | 2017-12-19 | 2017-12-19 | A kind of the clothes similarity determination method and its system of feature based extraction |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108052952A true CN108052952A (en) | 2018-05-18 |
Family
ID=62130160
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711377528.0A Pending CN108052952A (en) | 2017-12-19 | 2017-12-19 | A kind of the clothes similarity determination method and its system of feature based extraction |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108052952A (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109165293A (en) * | 2018-08-08 | 2019-01-08 | 上海宝尊电子商务有限公司 | A kind of expert data mask method and program towards fashion world |
| CN109820490A (en) * | 2018-09-04 | 2019-05-31 | 宓保宏 | A kind of region symmetry method for objectively evaluating based on infrared thermal imaging technique |
| CN110413874A (en) * | 2019-06-17 | 2019-11-05 | 浙江工业大学 | A kind of clothes recommended method based on dress ornament attributes match |
| CN110457329A (en) * | 2019-08-16 | 2019-11-15 | 第四范式（北京）技术有限公司 | A kind of method and device for realizing personalized recommendation |
| WO2019223302A1 (en) * | 2018-05-23 | 2019-11-28 | 山东大学 | Dress collocation method and system based on attention knowledge extraction, and storage medium |
| CN110610499A (en) * | 2019-08-29 | 2019-12-24 | 杭州光云科技股份有限公司 | Method for automatically cutting local detail picture in image |
| WO2020247960A1 (en) | 2019-06-07 | 2020-12-10 | Elc Management Llc | Method and apparatus for cosmetic product recommendation |
| CN112598487A (en) * | 2021-02-20 | 2021-04-02 | 汇正(广州)企业管理咨询有限公司 | Enterprise informatization management method and system based on artificial intelligence |
| WO2022111271A1 (en) * | 2020-11-24 | 2022-06-02 | 华为云计算技术有限公司 | Clothing standardization detection method and apparatus |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105447529A (en) * | 2015-12-30 | 2016-03-30 | 商汤集团有限公司 | Costume detection and attribute value identification method and system |
| CN106021603A (en) * | 2016-06-20 | 2016-10-12 | 昆明理工大学 | Garment image retrieval method based on segmentation and feature matching |
-
2017
- 2017-12-19 CN CN201711377528.0A patent/CN108052952A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105447529A (en) * | 2015-12-30 | 2016-03-30 | 商汤集团有限公司 | Costume detection and attribute value identification method and system |
| CN106021603A (en) * | 2016-06-20 | 2016-10-12 | 昆明理工大学 | Garment image retrieval method based on segmentation and feature matching |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2019223302A1 (en) * | 2018-05-23 | 2019-11-28 | 山东大学 | Dress collocation method and system based on attention knowledge extraction, and storage medium |
| CN109165293A (en) * | 2018-08-08 | 2019-01-08 | 上海宝尊电子商务有限公司 | A kind of expert data mask method and program towards fashion world |
| CN109820490A (en) * | 2018-09-04 | 2019-05-31 | 宓保宏 | A kind of region symmetry method for objectively evaluating based on infrared thermal imaging technique |
| EP3980963A4 (en) * | 2019-06-07 | 2023-05-03 | ELC Management LLC | Method and apparatus for cosmetic product recommendation |
| WO2020247960A1 (en) | 2019-06-07 | 2020-12-10 | Elc Management Llc | Method and apparatus for cosmetic product recommendation |
| CN110413874A (en) * | 2019-06-17 | 2019-11-05 | 浙江工业大学 | A kind of clothes recommended method based on dress ornament attributes match |
| CN110457329B (en) * | 2019-08-16 | 2022-05-06 | 第四范式（北京）技术有限公司 | Method and device for realizing personalized recommendation |
| CN110457329A (en) * | 2019-08-16 | 2019-11-15 | 第四范式（北京）技术有限公司 | A kind of method and device for realizing personalized recommendation |
| CN110610499B (en) * | 2019-08-29 | 2020-10-20 | 杭州光云科技股份有限公司 | Method for automatically cutting local detail picture in image |
| CN110610499A (en) * | 2019-08-29 | 2019-12-24 | 杭州光云科技股份有限公司 | Method for automatically cutting local detail picture in image |
| WO2022111271A1 (en) * | 2020-11-24 | 2022-06-02 | 华为云计算技术有限公司 | Clothing standardization detection method and apparatus |
| EP4239520A4 (en) * | 2020-11-24 | 2024-04-17 | Huawei Cloud Computing Tech Co Ltd | Clothing standardization detection method and apparatus |
| CN112598487B (en) * | 2021-02-20 | 2022-02-22 | 苏州希艾欧信息科技有限公司 | Enterprise informatization management method and system based on artificial intelligence |
| CN112598487A (en) * | 2021-02-20 | 2021-04-02 | 汇正(广州)企业管理咨询有限公司 | Enterprise informatization management method and system based on artificial intelligence |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108052952A (en) | A kind of the clothes similarity determination method and its system of feature based extraction | |
| CN106250874B (en) | Recognition methods and the device of a kind of dress ornament and carry-on articles | |
| CN109670591B (en) | Neural network training method and image matching method and device | |
| CN104978762B (en) | Clothes threedimensional model generation method and system | |
| CN105893936B (en) | A kind of Activity recognition method based on HOIRM and Local Feature Fusion | |
| CN105608456B (en) | A kind of multi-direction Method for text detection based on full convolutional network | |
| CN107944431B (en) | A kind of intelligent identification Method based on motion change | |
| CN104680141B (en) | Facial expression recognizing method and system based on moving cell layering | |
| CN108109055B (en) | Cross-scene clothing retrieval method based on image rendering | |
| CN106504064A (en) | Clothes classification based on depth convolutional neural networks recommends method and system with collocation | |
| CN106021603A (en) | Garment image retrieval method based on segmentation and feature matching | |
| CN106547880A (en) | A kind of various dimensions geographic scenes recognition methodss of fusion geographic area knowledge | |
| CN104850825A (en) | Facial image face score calculating method based on convolutional neural network | |
| CN107391703A (en) | The method for building up and system of image library, image library and image classification method | |
| CN106778852A (en) | A kind of picture material recognition methods for correcting erroneous judgement | |
| CN109614925A (en) | Dress ornament attribute recognition approach and device, electronic equipment, storage medium | |
| CN107785061A (en) | Autism-spectrum disorder with children mood ability interfering system | |
| CN106960181A (en) | A kind of pedestrian's attribute recognition approach based on RGBD data | |
| CN110598017B (en) | Self-learning-based commodity detail page generation method | |
| KR102211400B1 (en) | Apparatus and method for generating designs based on image and text | |
| CN109948709A (en) | A kind of multitask Attribute Recognition system of target object | |
| CN110047081A (en) | Example dividing method, device, equipment and the medium of chest x-ray image | |
| CN106897669A (en) | A kind of pedestrian based on consistent iteration various visual angles transfer learning discrimination method again | |
| CN106845513A (en) | Staff detector and method based on condition random forest | |
| CN107870992A (en) | Editable image of clothing searching method based on multichannel topic model |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180518 |